Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Enhanced vascular permeability occurs during acute inflammation and follows a paracellular pathway. Agonist-induced endothelial cell (EC) gap formation is often accompanied by changes in cytoskeletal architecture, such as actin polymerization, stress fiber formation, and isometric force development that is consistent with the involvement of a contractile mechanism in EC barrier dysfunction (1). Smooth-muscle (SM) cell contractile responsiveness is mediated initially by a rise in [Ca²⁺]_i, Ca²⁺ binding to calmodulin (CaM), and the resulting Ca²⁺/CaM-dependent activation of a myosin- and actin-binding myosin light chain kinase (MLCK). In both SM and nonmuscle tissues such as EC, MLCK preferentially phosphorylates regulatory myosin light chains (MLC₂₀) at Ser-19, resulting in the initiation of contractile activity generated by the movement of actin and myosin filaments past one another (5, 6). In permeabilized endothelium, adenosine triphosphate (ATP), Ca²⁺, MLCK, and CaM were absolute requirements for retraction and intercellular gap formation, which was accompanied by MLC phosphorylation (7, 8). In intact EC, thrombin-induced MLC phosphorylation at Ser-19 and diphosphorylation at Ser19/Thr-18 precedes rapid isometric force development in endothelium (1). Inhibition of MLCK activity abolishes thrombin-induced EC gap formation and permeability responses (4). Recently, we identified the presence of a previously undetected nonmuscle MLCK isoform in endothelium that is distinct from both the mammalian SM MLCK and the predicted avian nonmuscle MLCK (9). Molecular cloning of this enzyme in cultured endothelium revealed a protein with a predicted molecular weight of 211 kD (apparent molecular weight on sodium dodecyl sulfate-polyacrylamide gel electrophoresis [SDS-PAGE] is 214 kD) which, between amino acids #923-1914, retains all structural domains characteristic of SM MLCK. These regional motifs include an actin-binding region, an MLC-binding domain, a catalytic domain, a CaM-binding domain, and a highly conservative C-terminal domain, which may be an independently expressed protein known as telokin or kinase-related protein (10, 11). Amino acid analysis of EC MLCK primary structure revealed conserved Ser/Thr phosphorylation sites adjacent to the CaM-binding domain (amino acid #1740- 1759), which may be important for enzyme regulation (12, 13). Sequence analysis also identified, however, a novel N-terminal protein sequence (amino acid #1-922) that is not contained in mammalian SM MLCK (9, 14) but includes potential phosphorylation sites for multiple protein kinases, including cyclic adenosine monophosphate (cAMP)- dependent protein kinase A (PKA), Ca²⁺/CaM-dependent protein kinase II (CaM-PK II), and protein kinase C (PKC), enzymes that regulate SM MLCK activity in vitro (12, 13). Our studies have concluded that EC MLCK enzymatic activity, in addition to Ca²⁺/CaM sensitivity, may be regulated by cAMP-dependent MLCK phosphorylation (9). In addition, the novel N-terminal fragment of EC MLCK contained several unique structural motifs, including a fibroblast growth factor receptor-like domain and sites for tyrosine phosphorylation, which may be potentially important for Ca²⁺/CaM-independent MLCK regulation (9). In a prior report contained in this issue (15), we studied the expression of EC MLCK in multiple human and bovine EC tissues and demonstrated a remarkable similarity between the bovine and human EC MLCK isoforms. In the present studies we have utilized biochemical and morphologic techniques to characterize further the novel EC MLCK isoform in bovine endothelium. These studies confirm that EC MLCK represents a novel nonmuscle MLCK isoform that is not regulated solely by increases in intracellular Ca²⁺.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents

Endothelial cell cultures were maintained in DME media (Gibco, Chagrin Falls, OH) supplemented with 20% (vol/ vol) colostrum-free bovine serum (Irvine Scientific, Santa Ana, CA), 15 µg/ml endothelial cell growth supplement (Collaborative Research, Bedford, MA), 1% antibiotic and antimycotic solution (penicillin, 10,000 units/ml; streptomycin, 10 mg/ml; and amphotericin B, 25 µg/ml) (K.C. Biologicals, Lenexa, KS), and 0.1 mM nonessential amino acids (Gibco). Unless specified, reagents were obtained from Sigma Chemical Company (St. Louis, MO). Phosphate-buffered saline (PBS) and Hanks' balanced salt solution without phenol red were purchased from Gibco (Grand Island, NY). Bovine thrombin was obtained from Sigma. Antiactin monoclonal antibodies were purchased from Oncogene Science (Uniondale, NY), antirabbit myosin antibodies from Biochemical Technologies, Inc. (Stoughton, MA), and anticalmodulin antibody from UBI (Lake Placid, NY).

Bovine Pulmonary Artery Endothelial Cell (BPAEC) Culture

The BPAEC monolayers used in these studies have been previously characterized (4, 16). These cells were obtained frozen at 16 passages from American Type Culture Collection (Rockville, MD; CCL 209) and were utilized at passage 19-24. BPAEC were cultured in complete media and maintained at 37°C in a humidified atmosphere of 5% CO₂-95% air and grew to contact-inhibited monolayers with typical cobblestone morphology. Cells from each primary flask were detached with 0.05% trypsin, resuspended in fresh culture medium, and passaged to 75-cm² flasks for MLC phosphorylation experiments, to 35-mm dishes for histologic studies, to 60-mm dishes for ³²P-labeling experiments, or to 100-mm dishes for MLCK immunoprecipitation studies.

MLC Phosphorylation in Intact Endothelium

MLC₂₀ phosphorylation profiles were analyzed by urea PAGE as we have previously described (4). Briefly, confluent BPAEC in 75-cm² tissue flasks were harvested by scraping in 10% trichloroacetic acid (TCA) and 10 mM dithiothreitol. After centrifugation, the pellets were washed three times with diethyl ether and suspended in 6.7 M urea sample buffer, and approximately 75 µg protein per lane were run on a 10% polyacrylamide/40% glycerol gel to separate unphosphorylated MLC from the more rapidly migrating phosphorylated MLC. The proteins were then transferred to nitrocellulose (25 V for 90 min), and both phosphorylated and unphosphorylated MLC were detected by immunostaining overnight with an MLC₂₀-specific antibody (1:2,000 dilution) followed by 1 h treatment with peroxidase-conjugated secondary antibodies (1:3,000 dilution; Bio-Rad, Hercules, CA). The blot was scanned on a Bio-Rad densitometer and the percent maximal MLC phosphorylation was expressed by dividing the sum of twice the diphosphorylated MLC area and the monophosphorylated MLC area by the total of phosphorylated and unphosphorylated areas (i.e., maximum phosphorylation is 200%).

EC Detergent Fractionation

EC proteins were separated into detergent-soluble and -insoluble fractions as described previously (17, 18). Briefly, stimulated and nonstimulated BPAEC from 100-mm dishes were rinsed twice with 2 ml PBS at room temperature and incubated with 1.5 ml of extraction buffer (1% NP-40; 150 mM NaCl; 50 mM NaF; 0.5 mM orthovanadate; 28 mM -mercaptoethanol in 50 mM Tris HCl, pH 8.0), containing proteinase inhibitors (0.5 mM phenylmethylsulfonyl fluoride [PMSF], 0.1 mM tosyl-lysine chloromethyl ketone [TLCK], 0.1 mM leupeptin, 2 mM benzamidine) for 30 min at 4°C. Extracts were clarified by microcentrifugation and aliquots of supernatants (i.e., detergent-soluble cytosolic EC fractions) were used for MLCK immunoprecipitation under nondenaturing conditions as described subsequently. Insoluble proteins remaining on dishes (i.e., detergent-insoluble cytoskeletal EC fractions) were rinsed twice with ice-cold PBS, solubilized by scraping dishes in 3 ml of Laemmli sample buffer (19) supplemented with 20 mM NaF and 0.5 mM orthovanadate, and subjected to Western immunoblotting analysis as described subsequently.

Western Immunoblotting

Proteins were extracted from BPAEC preparations using SDS sample buffer as previously described (18, 19). Extracts were separated by 4 to 15% gradient SDS-PAGE (Bio Rad) gel electrophoresis, transferred to nitrocellulose (30 V, 18 h) (20) and reacted with SM MLCK polyclonal antibody (D119; a generous gift from Dr. P. Gallagher, Indiana University, Indianapolis, IN), as previously described (9). Immunoreactive proteins were detected using an enhanced chemiluminescent detection system according to the manufacturer's directions (Amersham, Little Chalfont, Buckinghamshire, UK). The relative intensities of the protein in the bands were quantified by scanning densitometry.

MLCK Immunoprecipitation

For immunoprecipitation under nondenaturing conditions, confluent EC from 100-mm dishes were rinsed once with M199 media and twice with PBS, then lysed for 5 min on ice with 300 µl NP-40 lysis buffer (1% NP-40, 20 mM 3-[N-morpholino]propanesulfonic acid [MOPS, pH 7.0], 50 mM MgCl₂, 10% glycerol, 0.5 mM ethyleneglycol-bis- [-aminoethyl ether]-N,N'-tetraacetic acid [EGTA]) including protease inhibitors (40 µg/ml aprotinin, 18 µg/ml L-[tosylamide 2-phenyl] ethyl chloromethyl ketone, 6 µg/ ml TLCK, 0.5 mM PMSF) and 28 mM -mercaptoethanol. The lysate was scraped, then microcentrifuged for 5 min (ISS 218; Integrated Separation System, Portsmouth, NH) at 4°C, and the supernatant was used for immunoprecipitation. Each sample was diluted with 700 µl washing buffer (0.1% NP-40; 50 mM MOPS, pH 7.0; 50 mM MgCl₂; 1 mM ethylenediaminetetraacetic acid [EDTA]) and incubated for 1 h with 1 µl anti-MLCK D119 antibodies at 4°C, followed by incubation for 1 h at 4°C with 50 µl 10% Pansorbin suspension consisting of formalin-hardened and heat-killed Cowan 1 strain Staphylococcus aureus cells (Calbiochem, La Jolla, CA). The immunoprecipitated complex was harvested by microcentrifugation, washed three times with washing buffer, and used for MLCK activity assay (see below) or resuspended in 200 µl of the Laemmli sample buffer (19) and heat-treated at 100°C for 5 min. Immunoprecipitated proteins were separated from the Pansorbin beads by microcentrifugation for 1 min and subjected to Western immunoblotting analysis (20) using specific antibodies to contractile proteins.

For immunoprecipitation under denaturing conditions, confluent EC monolayers in 60-mm tissue culture dishes were labeled with [³²P] orthophosphate (0.5 mCi/plate) for 2.5 h in phosphate-free DMEM (Sigma) without serum, followed by stimulation with either vehicle alone, phorbol myristate acetate (PMA, 100 nM), thrombin (100 nM), NaF (20 mM), or calyculin (10 nM). The stimuli were then removed and the monolayers were rinsed twice with 2 ml media, further rinsed with 2 ml PBS, and scraped into 100 µl SDS/denaturing stop solution (PBS, pH 7.4; 1 mM EDTA; 1 mM EGTA; 50 mM NaF; 10 mM NaPP; 0.2 mM orthovanadate; 1% SDS; 14 mM -mercaptoethanol). The homogenate was prepared by passing the cell suspension several times through a 16-gauge needle. Homogenates were heat-treated at 110°C for 5 min, diluted 1:10 with 900 µl PBS, and incubated with 50 µl of 10% Pansorbin suspension for 30 min at room temperature. Samples were clarified by microcentrifugation for 5 min and supernatants were incubated with 10 µl anti-MLCK antibodies (60 min at room temperature or overnight at 4°C), then with 50 µl of 10% Pansorbin suspension for 60 min at room temperature.

Immunocomplexes were pelleted by microcentrifugation for 5 min, washed three times with 1 ml PBS, boiled for 5 min in 100 µl of SDS sample buffer (19), separated from Pansorbin beads by microcentrifugation, and subjected to SDS electrophoresis (19). After electrophoresis, proteins were transferred to nitrocellulose membranes (20) and ³²P signals were detected by autoradiography. To identify the position of EC MLCK, membranes were subsequently stained with specific MLCK antibodies. The relative intensities of the ³²P-labeled MLCK were quantified by scanning densitometry.

Immunofluorescence

The fluorescent imaging of paracellular gap formation and of F-actin organization was performed on EC monolayers grown to confluence on glass coverslips. After treatment, cells were fixed by exchanging media with 5% paraformaldehyde; 50 mM phosphate; 75 mM NaCl; 25 mM Tris, pH 7, on ice for 10 min. Cells were thoroughly rinsed with buffer containing 150 mM NaCl and 50 mM Tris, pH 7.6, and then permeabilized by 3.5 min treatment with 0.2% Triton in rinse buffer. Cells were again rinsed three times, incubated at room temperature for 1 h with 1% bovine serum albumin (BSA) in rinse buffer, and then rinsed with 1 U/ml rhodamine phalloidin (Molecular Probes, Eugene, OR) to identify F-actin. Time-dependent changes in intracellular distribution of the actin cytoskeleton after 100-nM thrombin challenge were analyzed on a Zeiss Axioplan Fluorescent Microscope with an MC100 camera as we previously described (21). To study colocalization of actin and MLCK, the fixed, permeabilized cells were exposed overnight at 4°C to a 1:50 dilution of anti-EC MLCK antibody (V-368) in BSA buffer. This antibody was generated against the peptide GEERKRP present in the unique N-terminal part of EC MLCK (9). After rinsing to remove unbound primary antibody, cells were incubated for 1 h at room temperature with labeled secondary antibody (30 µg/ml fluorescein isothiocyanate [FITC]-conjugated donkey antirabbit IgG; Jackson Co., West Grove, PA) and rhodamine phalloidin. Cells were examined using a ×60 oil objective with the Bio-Rad MRC 1024 confocal microscope and excitation with Ar/Kr laser at 568 nm excitation/ 598 emission for rhodamine and 488 excitation/522 emission for FITC at a 3-µm aperture. Data were collected for 7 to 17 planar sections at 0.5-µm intervals by Bio-Rad LaserSharp acquisition software, processed by MetaMorph Imaging software (Universal, West Chester, PA), and printed on a thermal dye diffusion printer (Kodak, Rochester, NY). EC monolayers that were not exposed to primary antibody showed no staining with the secondary antibody.

Determination of MLCK Activity

MLCK activity was determined in nondenaturing immunoprecipitates prepared from BPAEC homogenates with D119 anti-MLCK antibodies as we have previously described (9). Immunocomplexes were resuspended in 110 µl of 50 mM MOPS (pH 7.4), 10 mM magnesium acetate, 1 mg/ml BSA, and 8 mM -mercaptoethanol, and preincubated with or without the specific MLCK inhibitor KT5926 (10 µM) for 15 min at 25°C. Kinase activity in MLCK immunoprecipitates was measured using MLC obtained from bovine muscle (1 mg/ml; Sigma) as a substrate in a buffer consisting of 50 mM MOPS, pH 7.4; 10 mM magnesium acetate; 1 mg/ml BSA; 1 µM CaM; 0.1 mM -³²P ATP (1 Ci/mMol); and 0.3 mM CaCl₂ for 30 min at 25°C. Reaction was stopped by pipetting aliquots onto Whatman 3MM filters and immediately rinsing with ice-cold 10% TCA and 2% sodium pyrophosphate (wt/vol), followed by two 15-min washings with 95% ethanol. Finally, filters were rinsed in ethyl ether, dried, and counted by liquid scintillation counting. Specific MLCK activity was measured by subtracting the level of kinase activity insensitive to the specific MLCK inhibitor KT5926.

CaM Binding

MLCK immunoprecipitates were prepared under nondenaturing conditions and transferred to nitrocellulose as described previously. Nonspecific protein binding sites were blocked for 30 min at room temperature using 5% nonfat dry milk in buffer A (22 mM Tris HCl, pH 7.6; 150 mM NaCl). Nitrocellulose membranes were washed 3 times for 10 min with buffer A and incubated with biotinylated CaM (Biochemical Technologies, Inc.) at a concentration of 2 µg/ ml in buffer A with 1 mM CaCl₂ or 2 mM EGTA (30 min at room temperature). Unbound probe was removed by washing as above and membranes were treated with peroxidase-labeled avidine (Sigma) at a final concentration of 2.5 µg/ml in buffer A for 20 min followed by washing and ECL detection (Amersham). In parallel experiments, MLCK was detected in immunoprecipitates by Western immunoblotting with MLCK-specific antibodies and compared with the CaM-binding studies described.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Functional Characterization of EC MLCK in Bovine Endothelium

In a prior report contained in this issue (15), we demonstrated significant molecular and structural similarity between the bovine EC MLCK isoform and the previously cloned human EC MLCK (9). To characterize EC MLCK further, we next examined the Ca²⁺/CaM binding properties and functional characteristics of bovine EC MLCK. As depicted in Figure 1, the 214-kD EC MLCK isoform bound biotinylated CaM only in the presence of Ca²⁺, verifying the classic Ca²⁺-dependent, CaM-binding feature of this enzymatic family. This observation is also consistent with the Ca²⁺/CaM-dependency of EC MLCK activation we have previously noted in bovine and human EC (4). We next analyzed proteins coimmunoprecipitated with EC MLCK antisera under nondenaturing conditions and identified EC MLCK, myosin heavy and light chains, actin, and calmodulin (Figure 2), suggesting that these contractile proteins exist in a functional complex. Western blotting of MLCK immunoprecipitates performed with MAP kinase-specific antibodies revealed the absence of MAP kinase in MLCK immunoprecipitates (data not shown), indicating the specificity of the results obtained. Densitometric quantitation of proteins present in either the MLCK immunoprecipitates or supernatants revealed that nearly 100% of myosin, ~ 50% of CaM, and ~ 20% of actin complexed with EC MLCK under these conditions, suggesting that EC MLCK binds more tightly with cellular myosin than with actin under unstimulated conditions.

View larger version (31K): [in this window] [in a new window]   View larger version (25K): [in this window] [in a new window]   Figure 1.   Binding of biotinylated CaM by EC MLCK. (A) BPAEC proteins present in MLCK immunoprecipitates (IP) and supernatants (SN) were subjected to gradient SDS-PAGE separation and Western immunoblotting using either anti-SM MLCK D119 antibodies (left) or biotinylated CaM (2 µg/ml) in the presence of 1 mM CaCl2 (right). (B) BPAEC proteins present in MLCK immunoprecipitates were subjected to 7.5% SDS-PAGE separation and Western immunoblotting using either biotinylated CaM (2 µg/ml) in the presence of 2 mM EGTA (EGTA/CaM), or 1 mM CaCl2 (calcium/CaM), or anti-MLCK D119 antibodies (left lane). In each panel, each lane represents equivalent amount of total proteins as normalized by Coomassie-stained gel quantitation. Positions of the molecular weight markers and EC MLCK are indicated. The high molecular-weight EC MLCK, present only in the MLCK immunoprecipitated fraction, binds biotinylated CaM in a Ca2+-dependent manner.

View larger version (33K): [in this window] [in a new window]   Figure 2.   Immunoprecipitation of EC MLCK under nondenaturing conditions. Anti-MLCK immunoprecipitates from BPAEC extracts were prepared under nondenaturing conditions as described in MATERIALS AND METHODS and subjected to gradient SDS-PAGE (4 to 15%). This was followed by Western immunoblotting with specific anti-SM MLCK D119 antibodies (lane 1), anti-actin monoclonal antibodies (lane 2), anti-MLC-specific antibodies (lane 3), anti-platelet myosin antibodies (lane 4), and anti-CaM antibodies (lane 5). Positions of corresponding contractile proteins and rabbit IgG are indicated by arrows. Under nondenaturing conditions, EC MLCK appears to immunoprecipitate in a functional complex with several key contractile proteins.

Effect of Thrombin on EC MLCK Activity

We have previously shown that EC activation by agonists such as thrombin leads to rapid and significant increases in MLC phosphorylation, activation of the EC contractile apparatus, and finally EC barrier dysfunction (4). It is well known that the level of MLC phosphorylation actually reflects a balance between MLCK and myosin-associated phosphatase activities (22). To estimate directly the contribution of MLCK activation to agonist-mediated contractile responses in endothelium, we recently developed a MLCK activity assay in the nondenaturing immunoprecipitates using total MLC from bovine muscle (Sigma) as a substrate. Using this assay we have shown that specific MLCK inhibitors, KT5926 and ML-7, significantly decrease EC MLCK activity in the immunoprecipitates (9). To link the level of MLC phosphorylation with the extent of MLCK activity, we measured both parameters at specific time points after thrombin-induced EC activation. Figure 3 demonstrates that 100 nM thrombin rapidly increases MLC phosphorylation (maximal at 2 min with a greater than 2-fold increase). The extent of MLC phosphorylation was highly correlated with the increase in MLCK activity (~ 2-fold increase; Figure 3) and with the rapid Ca²⁺ transient initiated by thrombin (23). MLCK activity (Figure 3) and intracellular Ca²⁺ (23) decline to near basal level by 10 min, whereas the level of MLC phosphorylation declined more slowly and was sustained above basal levels even after 60 min of stimulation. We speculate that the delayed decline in MLC phosphorylation reflects thrombin-induced inhibition of myosin-associated phosphatase activity that we have noted as early as 10 min after agonist stimulation (24). Therefore, these data suggest that the initial increases in MLC phosphorylation after thrombin can be attributed to Ca²⁺-dependent increases in MLCK activity.

View larger version (33K): [in this window] [in a new window]   Figure 3.   Temporal activation of BPAEC MLCK activity by thrombin. Confluent EC monolayers were rinsed twice with M199 media to remove serum and were incubated in M199 in lieu of bicarbonate. Cells were treated with vehicle (M199) or 100 nM thrombin for indicated time periods. MLCK activity in nondenaturing MLCK immunoprecipitates was measured using MLC as a substrate as described in MATERIALS AND METHODS. MLC phosphorylation was monitored by urea gel electrophoresis followed by immunoblotting with anti-MLC antibodies (4) and quantitation by laser densitometry. Data are reported as the means ± SE (n = 4-14). *Significant difference (P < 0.05) compared with control; #significant difference compared with the maximal level of thrombin-induced EC responses.

Effect of Thrombin on EC MLCK Cellular Distribution

Figure 4 demonstrates that thrombin-mediated EC activation is accompanied by rapid (2 min) redistribution of EC MLCK to the detergent-insoluble cytoskeleton with a concomitant decrease in the level of immunoprecipitable MLCK from the detergent-soluble supernatant fraction. This thrombin-mediated MLCK translocation is consistent with the contractile phenotype previously noted after thrombin (1, 16, 21), which includes rapid and prominent cytoskeletal rearrangements, such as an increase in F-actin content, time-dependent reorganization of the actin cytoskeleton, and dissolution of the dense peripheral actin band (Figure 5). Confocal immunofluorescence experiments revealed near-complete absence of actin and MLCK colocalization in unstimulated cells but a time-dependent increase in the colocalization of EC MLCK with F-actin after thrombin stimulation (Figure 6). The dramatic MLCK cellular translocation coincides well with maximal MLCK activity and extent of MLC phosphorylation (Figure 3) and precedes thrombin-induced actin redistribution that was evident after 10 min of activation (Figure 4). These studies suggest that in addition to increased Ca²⁺/CaM availability and subsequent CaM binding, thrombin-induced MLCK activation may be highly dependent upon the extent of MLCK-cytoskeletal protein interaction. Consistent with this hypothesis, we have recently shown that disruption of the endothelial cell microtubule system dramatically increases EC MLCK-mediated MLC phosphorylation (A. D. Verin and J. G. N. Garcia, unpublished data).

View larger version (27K): [in this window] [in a new window]   Figure 4.   Translocation of EC MLCK to the cytoskeleton by thrombin. A shows an immunoblot of detergent-insoluble BPAEC fractions stained with anti-MLCK antibodies (upper photograph) and restained with antiactin antibodies obtained from control () and thrombin-stimulated (+) BPAEC (100 nM). B shows an MLCK immunoblot of detergent-soluble BPAEC fractions after MLCK immunoprecipitation from thrombin-treated (+) (100 nM, 2 min) or control () cells. A dramatic increase in the level of MLCK detected in the detergent-insoluble cytoskeletal fraction was observed after thrombin stimulation and was accompanied by a corresponding decrease in EC MLCK detected in the detergent-soluble cytosolic fraction, indicating that thrombin stimulates MLCK translocation to the cytoskeleton.

View larger version (110K): [in this window] [in a new window]   Figure 5.   Thrombin-induced EC actin polymerization. BPAEC cultured on glass coverslips were exposed to either vehicle (A) or 100 nM thrombin (B) for 10 min and then fixed with 5% paraformaldehyde. Cells were stained for F-actin with rhodamine phalloidin as described in MATERIALS AND METHODS. Both cell monolayers were photographed with an 8-s exposure with a ×60 objective. Control cells (A) show the normal fine weblike pattern of actin with prominent dense peripheral bands at the cell/cell borders (white arrow) and close, continuous apposition of cells. In contrast, the brief exposure to thrombin (B) results in a prominent increase in F-actin content and increased actin fiber thickness consistent with rapid polymerization. Moreover, there is marked dissolution of the dense peripheral bands and appearance of numerous small intercellular gaps (open black arrow).

View larger version (41K): [in this window] [in a new window]   Figure 6.   Effect of thrombin stimulation on EC MLCK intracellular localization. BPAEC were treated with either vehicle (A), 100 nM thrombin for 2 min (B), or 100 nM thrombin for 10 min (C), and fixed with paraformaldehyde as described in MATERIALS AND METHODS. Fixed, permeabilized cells were imaged by confocal microscopy after immunostaining with primary antibody to MLCK (V-368), post-staining with FITC-antirabbit IgG (green), and staining with rhodamine phalloidin (red) to assess F-actin binding. As noted in B and C, thrombin induced a rapid increase in F-actin staining and a time-dependent increase in the colocalization of MLCK and actin, as indicated by the shift in MLCK with staining from predominantly green in the control cells to yellow (mixture of the red and green fluorescence) after thrombin challenge.

Regulation of EC MLCK by Ser/Thr Phosphorylation

Protein phosphorylation/dephosphorylation are important mechanisms that regulate enzymatic activity and specific protein translocation to the detergent-insoluble cytoskeleton (25). Analysis of the predicted amino acid sequence of the novel human EC MLCK isoform (9) revealed multiple potential phosphorylation sites for a variety of protein kinases (Table 1). Importantly, potential phosphorylation sites on Ser/Thr residues were conserved within or adjacent to the CaM-binding domain, including identical PKA consensus sequences for phosphorylation of serine residue #1773, a site reported as potentially important in the regulation of SM MLCK activity (12, 26, 27). We next sought to extend our earlier observation (9) that phosphorylation of the EC MLCK isoform may represent a potential signal for regulating enzymatic activity. Intact EC monolayers were loaded with ³²P-orthophosphate and treated with either vehicle control or specific agonists, followed by immunoprecipitation of EC MLCK under denaturing conditions. Although endogenous MLCK phosphorylation was noted in vehicle-treated controls, despite numerous potential PKC phosphorylation sites (Table 1), the extent of EC MLCK phosphorylation did not significantly increase at the time points analyzed beyond basal levels with either thrombin or PMA treatment (Table 2), two modalities known to produce substantial PKC activation (28, 29). However, levels of MLCK phosphorylation were significantly increased (~ 1.4- to 2-fold) by inhibition of Ser/Thr phosphatase activity (Table 2, Figure 7) accomplished with either 20 mM NaF, a potent nonspecific phosphatase inhibitor (30), or 10 nM calyculin, an intervention we have shown to inhibit type 1 and 2A EC phosphatase activity by > 95% (18). The extent of MLCK phosphorylation was inversely correlated with MLCK activity when measured in the presence of 0.3 mM Ca²⁺ in the assay buffer, a condition that allows full activation of the enzyme (Figure 7). These data are consistent with endogenous Ser/Thr phosphorylation of the novel EC MLCK isoform, and support the hypothesis that in addition to Ca²⁺/CaM availability, the extent of MLCK activity is linked to alterations in the phosphorylation status of the enzyme.

View larger version (40K): [in this window] [in a new window]   Figure 7.   Effect of phosphatase inhibition on EC MLCK phosphorylation and activity. For determination of EC MLCK phosphorylation, BPAEC monolayers loaded with 32P to radiolabel ATP pools and with vehicle (DMEM for NaF, 0.1% dimethyl sulfoxide for calyculin) or phosphatase inhibitors (20 mM NaF or 10 nM calyculin) for 1 h at 37°C. MLCK phosphorylation was determined after immunoprecipitation with anti-MLCK D119 antibodies under denaturing conditions, and the autoradiogram was subjected to quantitative scanning densitometry. EC MLCK activity was determined in nondenaturing immunoprecipitates using MLC as a substrate and expressed as a percentage change from control. *Significant difference (P < 0.05) from control level of MLCK activity and phosphorylation. Inset shows an autoradiogram of MLCK immunoprecipitates from NaF-treated or control cells loaded with 32P.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The results from this study indicate that the novel high molecular-weight nonmuscle MLCK isoform present in bovine and human endothelium share many biochemical properties with the extensively characterized SM MLCK isoform. Bovine EC MLCK uses MLC as a substrate, requires cytosolic Ca²⁺ availability (4), and binds CaM in a Ca²⁺-dependent manner (Figure 1) with its activity reduced by specific MLCK inhibitors (4, 9). Furthermore, EC MLCK complexes with contractile proteins such as myosin and actin (Figure 2), and its enzymatic activity is increased by agonists that increase the intracellular concentration of ionized calcium ([Ca²⁺]_i), such as the coagulation protease thrombin (Figure 3). Unlike SM MLCK, however, an increase in [Ca²⁺]_i alone is insufficient to activate EC MLCK (4, 31).

We previously concluded that EC MLCK is a key participant in agonist-induced EC contractility and barrier dysfunction (4). To study the mechanisms of regulation of thrombin-mediated EC contractility, we examined the effect of thrombin on EC MLCK activity, MLCK phosphorylation status, and cellular localization of the EC MLCK isoform. EC MLCK activity rapidly increased after thrombin stimulation, with increased MLC phosphorylation following transient increases in intracellular Ca²⁺ (4, 23). The decline in [Ca²⁺]_i was temporally linked to a rapid decrease in MLCK enzymatic activity and a decrease in MLC phosphorylation. These findings are consistent with our previous reports that thrombin activates MLCK via Ca²⁺-dependent mechanisms in endothelium (4). We also observed that the increase in MLCK activity and MLC phosphorylation seen after thrombin is associated with rapid MLCK translocation to the actin-linked, detergent-insoluble cytoskeleton. Although the mechanism underlying this process is unclear, we believe this can be attributed to cytoskeletal rearrangements stimulated by thrombin (20, 32). Changes in cytoskeletal architecture initiated by microtubule disruption have been noted to lead to rapid increases in MLC phosphorylation in fibroblasts (33) and in EC (A. D. Verin and J. G. N. Garcia, unpublished observations), which correlate with increased contraction (33). A number of signaling molecules, including p60^c-src, phospholipase C, and protein tyrosine phosphatases, relocate to the cytoskeleton during thrombin-induced platelet activation (34), and this process appears to be dependent on integrin-dependent cytoskeletal reorganization (36, 38). Very recent studies have reported that a high molecular-weight MLCK isoform interacts with integrins (40, 41), suggesting that the thrombin-induced EC MLCK translocation to the cytoskeleton may likewise be integrin- mediated. Coimmunoprecipitation experiments (Figure 2) indicated that, similar to SM MLCK (42), EC MLCK possesses F-actin binding ability that may be related to amino acids #923-964, which are highly homologous with the putative actin-binding region described for SM MLCK. We speculate that EC MLCK translocation to the cytoskeleton reflects increased actin-binding affinity during thrombin-induced EC cytoskeletal rearrangement. In vitro motility assay experiments have shown that the addition of purified SM MLCK to purified actin in the presence of Ca²⁺/CaM increases the velocity of actin filament movement toward phosphorylated myosin (43). This stimulating effect of MLCK was not attributable to MLC kinase activity but could be explained by its actin-binding activity (43). In light of these findings, we hypothesize that additional undefined regulatory elements, aside from Ca²⁺/CaM availability, may be involved in MLCK-mediated regulation of contractility in specific cell types.

The post-translational modification of MLCK via phosphorylation represents a potentially important mechanism of MLCK regulation. We have recently cloned MLCK in endothelium (9), and although sequence analysis revealed substantial (> 95%) homology to the coding region of SM MLCK between amino acids #923-1914, a novel N-terminal sequence (amino acids #1-922) (9) was identified that is not contained in the open reading frame of SM MLCK (14). The unique N-terminal fragment included several potential phosphorylation sites for protein kinases, such as PKA, PKC, and CaM-PK II (Table 1). We speculate that the existence of unique phosphorylation sites in EC MLCK may represent sites of enzymatic regulation of this isoform in endothelium. In this regard, we demonstrated that inhibition of Ser/Thr phosphatase activity markedly increased the level of EC MLCK phosphorylation, a finding that strongly and temporally correlated with significant decreases in MLCK activity in immunoprecipitates (Figure 6). In SM MLCK, phosphorylation at site A adjacent to the CaM-binding domain (corresponding to amino acid #1749-1759 in EC MLCK) decreases MLCK affinity for CaM and the Ca²⁺ sensitivity of in vitro enzyme activity when measured at a constant concentration of CaM (13). The functional significance of SM MLCK phosphorylation at sites distinct from site A is poorly understood (44). At least six Ser/Thr enzymes, including PKC, PKA, cyclic guanidine monophosphate (cGMP)-dependent protein kinase, mitogen-activated protein kinase, cyclin-dependent kinase I, and CaM-PK II, are able to phosphorylate purified SM MLCK in vitro (12, 27, 44). In vivo activation of PKC, PKA, and CaM-PK II leads to significant ³²P incorporation in SM MLCK (13, 26, 48). However, only stimulation of CaM-PK II or pretreatment of tracheal SM with carbachol or KCl (both of which increase cytosolic [Ca²⁺] and initiate contraction), resulted in extensive phosphorylation at site A (13, 49). Pretreatment of SM cells with the Ca²⁺ ionophore ionomycin resulted in rapid [Ca²⁺]_i increases and MLC phosphorylation followed by SM MLCK phosphorylation catalyzed by CaM-PK II (13). We have recently shown that activation of endothelial cell cAMP-dependent PKA by cholera toxin markedly enhanced MLCK phosphorylation while reducing immunoprecipitated EC MLCK activity by 4-fold (9). However, in contrast to SM MLCK, we have recently reported that EC treatment with ionomycin dramatically decreased MLC phosphorylation while increasing Ser/Thr phosphatase activity (31). Pretreatment with the CaM-PK II inhibitor KN93 decreased ionomycin-induced CaM-PK II activity, but did not significantly affect ionomycin-induced MLC dephosphorylation (A. D. Verin, Shu Shi, and J. G. N. Garcia, unpublished data), whereas similar CaM-PK II inhibition in SM cells potentiated MLC phosphorylation in response to ionomycin (49). Surprisingly, despite an increase in [Ca²⁺]_i, EC monolayers challenged with thrombin (15 min) to stimulate EC contraction (4) did not significantly alter ³²P incorporation in EC MLCK. These data indicate that regulation of MLCK activity is isoform-specific in the two tissues and highlights the need for site-specific information regarding phosphorylation of specific Ser/Thr and Tyr residues.

In summary, immunologic and biochemical data provided in this report have characterized the novel high molecular-weight MLCK isoform present in bovine endothelium. Despite significant functional similarities with SM MLCK, the bovine EC MLCK isoform possesses unique regulatory properties that are not solely dependent on Ca²⁺/CaM availability but are significantly determined by both MLCK interaction with cytoskeletal proteins and the phosphorylation status of MLCK. Further studies that fully evaluate the significance of site-specific phosphorylation in the regulation of this important kinase will enhance our understanding of EC MLCK function and lend insight into the pathogenesis of high-permeability pulmonary edema.